molecular biologist guide to proteo mics

7/30/2019 Molecular Biologist Guide to Proteo Mics

1/25

MICROBIOLOGY AND MOLECULAR BIOLOGY REVIEWS, Mar. 2002, p. 3963 Vol. 66, No. 11092-2172/02/$04.000 DOI: 10.1128/MMBR.66.1.3963.2002Copyright 2002, American Society for Microbiology. All Rights Reserved.

Molecular Biologists Guide to ProteomicsPaul R. Graves1 and Timothy A. J. Haystead1,2*

Department of Pharmacology and Cancer Biology, Duke University,1 and Serenex Inc.,2 Durham, North Carolina 27710

INTRODUCTION ..................... ..................... ..................... .................... ..................... ...................... ..................... ......40Definitions..................................................................................................................................................................40Proteomics Origins .................. ..................... ..................... ..................... ..................... ..................... ..................... ...40Genome Information ................................................................................................................................................41

Why Proteomics?...................... ..................... ..................... ..................... ..................... ..................... ..................... ...41Annotation of the genome.......... ..................... ..................... ..................... .................... ...................... .................41Protein expression studies...................................................................................................................................41Protein function ....................................................................................................................................................42Protein modifications ...........................................................................................................................................42Protein localization and compartmentalization ...............................................................................................42Protein-protein interactions ................... ..................... ..................... ..................... ..................... ..................... ....42

Types of Proteomics..................................................................................................................................................42Protein expression proteomics ................... ..................... ..................... ..................... ..................... .....................42

Structural proteomics...........................................................................................................................................42Functional proteomics..........................................................................................................................................42

TECHNOLOGY OF PROTEOMICS .................... ..................... ..................... ..................... ..................... .................43Separation and Isolation of Proteins.....................................................................................................................43

One- and two-dimensional gel electrophoresis.................................................................................................43Alternatives to electrophoresis......... .................... ..................... ..................... ..................... ..................... ...........44

Acquisition of Protein Structure Information .................. .................... ..................... ..................... ......................45Edman sequencing................................................................................................................................................45Mass spectrometry................................................................................................................................................45

(i) Sample preparation.....................................................................................................................................45(ii) Sample ionization.......................................................................................................................................45(iii) Mass analysis ............................................................................................................................................46(iv) Types of mass spectrometers .................... ..................... ..................... ..................... ..................... ...........48(v) Peptide fragmentation................................................................................................................................48(vi) Our approach to mass spectrometry ..................... ..................... ..................... ..................... ..................49

Database Utilization.................................................................................................................................................50Peptide mass fingerprinting database searching .............................................................................................50

Amino acid sequence database searching .......... ..................... ..................... ..................... ..................... ...........50De novo peptide sequence information..............................................................................................................52Uninterpreted MS/MS data searching...............................................................................................................52

PROTEOMICS APPLICATIONS...............................................................................................................................53Characterization of Protein Complexes.................................................................................................................53Protein Expression Profiling ................... ..................... ..................... ..................... ..................... ..................... .......53

Expression profiling by two-dimensional electrophoresis...............................................................................53Isotope-coded affinity tags ................... ..................... ..................... ..................... ..................... ..................... .......53Protein arrays........................................................................................................................................................55

Proteomics Approach to Protein Phosphorylation...............................................................................................55Phosphoprotein enrichment ................................................................................................................................55Phosphorylation site determination by Edman degradation .................... ..................... ..................... ............55

Phosphorylation site determination by mass spectrometry............................................................................57(i) Phosphopeptide sequencing by MS/MS .................. ..................... ..................... ..................... ..................57(ii) Analysis of phosphopeptides by MALDI-TOF .................... .................... ..................... ..................... .....58

Yeast Genomics and Proteomics.................... ..................... ..................... ..................... ..................... .....................58Proteome Mining ......................................................................................................................................................58Challenges for Proteomics.......................................................................................................................................60

ACKNOWLEDGMENTS .......................... ..................... ..................... ..................... ..................... ..................... ..........60REFERENCES ................... ..................... ..................... ..................... ..................... ..................... ..................... .............60

* Corresponding author. Mailing address: Department of Pharma-cology and Cancer Biology, Duke University, Research Dr., C118LSRC, Durham, NC 27710. Phone: (919) 613-8606. Fax: (919) 668-0977. E-mail: [email protected].

39


2/25

INTRODUCTION

This review is intended to give the molecular biologist arudimentary understanding of the technologies behind pro-teomics and their application to address biological questions.Entry of our laboratory into proteomics 5 years ago was drivenby a need to define a complex mixture of proteins (36 pro-teins) we had affinity isolated that bound specifically to thecatalytic subunit of protein phosphatase 1 (PP-1, a serine/threonine protein phosphatase that regulates multiple dephos-phorylation events in cells) (26). We were faced with the taskof trying to understand the significance of these proteins, andthe only obvious way to begin to do this was to identify them bysequencing. We then bought an Applied Biosystems auto-mated Edman sequencer (not having the budget for a massspectrometer at the time). Since the majority of intact eukary-otic proteins are not immediately accessible to Edman se-quencing due to posttranslational N-terminal modifications,

we invented mixed-peptide sequencing (38). This method, de-scribed in detail later, essentially enables internal peptide se-

quence information to be derived from proteins electroblottedonto hydrophobic membranes. Using the mixed-peptide se-quencing strategy, we identified all 36 proteins in about a week.The mixture contained at least two known PP-1 regulatorysubunits, but most were identified in the expressed sequencetag or unannotated DNA databases and were novel proteins ofunknown function. Since that time, we have been using variousmolecular biological approaches to determine the functions ofsome of these proteins. Herein lies the lesson of proteomics.Identifying long lists of potentially interesting proteins oftengenerates more questions than it seeks to answer.

Despite learning this obvious lesson, our early sequencingexperiences were an epiphany that has subsequently altered

our whole scientific strategy for probing protein function incells. The sequencing of the 36 proteins has opened new ave-nues to further explore the functions of PP-1 in intact cells.Because of increased sensitivity, our approaches now routinelyuse state-of-the-art mass spectrometry (MS) techniques. How-ever, rather than using proteomics to simply characterize largenumbers of proteins in complex mixtures, we see the realapplication of this technology as a tool to enhance the power ofexisting approaches currently used by the modern molecularbiologist such as classical yeast and mouse genetics, tissueculture, protein expression systems, and site-directed mutagen-esis. Importantly, the one message we would want the reader totake away from reading this review is that one should always let

the biological question in mind drive the application of pro-teomics rather than simply engaging in an orgy of proteinsequencing. From our experiences, we believe that if the ap-propriate controls are performed, proteomics is an extremelypowerful approach for addressing important physiologicalquestions. One should always design experiments to define aselected number of relevant proteins in the mixture of interest.Examples of such experiments that we routinely perform in-clude defining early phosphorylation events in complex proteinmixtures after hormone treatment of intact cells or comparingpatterns of protein derived from a stimulated versus nonstimu-lated cell in an affinity pull-down experiment. Only the proteinsthat were specifically phosphorylated or bound in response tothe stimulus are sequenced in the complex mixtures. Sequenc-

ing proteins that are regulated then has a meaningful outcomeand directs all subsequent biological investigation.

Definitions

The term proteomics was first coined in 1995 and was

defined as the large-scale characterization of the entire proteincomplement of a cell line, tissue, or organism (13, 163, 167).Today, two definitions of proteomics are encountered. Thefirst is the more classical definition, restricting the large-scaleanalysis of gene products to studies involving only proteins.The second and more inclusive definition combines proteinstudies with analyses that have a genetic readout such asmRNA analysis, genomics, and the yeast two-hybrid analysis(123). However, the goal of proteomics remains the same, i.e.,to obtain a more global and integrated view of biology bystudying all the proteins of a cell rather than each one indi-

vidually.Using the more inclusive definition of proteomics, many

different areas of study are now grouped under the rubric of

proteomics (Fig. 1). These include protein-protein interactionstudies, protein modifications, protein function, and proteinlocalization studies to name a few. The aim of proteomics isnot only to identify all the proteins in a cell but also to createa complete three-dimensional (3-D) map of the cell indicating

where proteins are located. These ambitious goals will cer-tainly require the involvement of a large number of differentdisciplines such as molecular biology, biochemistry, and bioin-formatics. It is likely that in bioinformatics alone, more pow-erful computers will have to be devised to organize the im-mense amount of information generated from theseendeavors.

In the quest to characterize the proteome of a given cell or

organism, it should be remembered that the proteome is dy-namic. The proteome of a cell will reflect the immediate en-

vironment in which it is studied. In response to internal orexternal cues, proteins can be modified by posttranslationalmodifications, undergo translocations within the cell, or besynthesized or degraded. Thus, examination of the proteomeof a cell is like taking a snapshot of the protein environmentat any given time. Considering all the possibilities, it is likelythat any given genome can potentially give rise to an infinitenumber of proteomes.

Proteomics Origins

The first protein studies that can be called proteomics beganin 1975 with the introduction of the two-dimensional gel byOFarrell (119), Klose (87), and Scheele (140), who beganmapping proteins from Escherichia coli, mouse, and guinea pig,respectively. Although many proteins could be separated and

visualized, they could not be identified. Despite these limita-tions, shortly thereafter a large-scale analysis of all humanproteins was proposed. The goal of this project, termed thehuman protein index, was to use two-dimensional protein elec-trophoresis (2-DE) and other methods to catalog all humanproteins (14). However, lack of funding and technical limita-tions prevented this project from continuing.

Although the development of 2-DE was a major step for-ward, the science of proteomics would have to wait until the

40 GRAVES AND HAYSTEAD MICROBIOL. MOL. BIOL. REV.


3/25

proteins displayed by 2-DE could be identified. One problemthat had to be overcome was the lack of sensitive protein-sequencing technology. Improving sensitivity was critical forsuccess because biological samples are often limiting and bothone-dimensional (1-D) and two-dimensional (2-D) gels havelimits in protein loading capacity. The first major technology toemerge for the identification of proteins was the sequencing ofproteins by Edman degradation (45). A major breakthrough

was the development of microsequencing techniques for elec-

troblotted proteins (68). This technique was used for theidentification of proteins from 2-D gels to create the first 2-Ddatabases (31). Improvements in microsequencing technologyresulted in increased sensitivity of Edman sequencing in the1990s to high-picomole amounts (6).

One of the most important developments in protein identi-fication has been the development of MS technology (11). Inthe last decade, the sensitivity of analysis and accuracy ofresults for protein identification by MS have increased by sev-eral orders of magnitude (11, 123). It is now estimated thatproteins in the femtomolar range can be identified in gels.Because MS is more sensitive, can tolerate protein mixtures,and is amenable to high-throughput operations, it has essen-

tially replaced Edman sequencing as the protein identificationtool of choice.

Genome Information

The growth of proteomics is a direct result of advances madein large-scale nucleotide sequencing of expressed sequencetags and genomic DNA. Without this information, proteinscould not be identified even with the improvements made inMS. Protein identification (by MS or Edman sequencing) relieson the presence of some form of database for the given organ-ism (122, 146). The majority of DNA and protein sequenceinformation has accumulated within the last 5 to 10 years (23).In 1995, the first complete genome of an organism was se-

quenced, that of Haemophilus influenzae (56). At the time ofthis writing, the sequencing of the genomes of 45 microorgan-isms has been completed and that of 170 more is under way(http://www.tiger.org/tdb/mdb/mdbcomplete.html). To date,five eukaryotic genomes have been completed: Arabidopsisthaliana (154), Saccharomyces cerevisiae (58), Schizosaccharo-myces pombe (128), Caenorhabditis elegans (1), and Drosophilamelanogaster(3, 113, 138). In addition, the rice (105), mouse(178a), and human (93, 161) genomes are near completion.

Why Proteomics?

Many types of information cannot be obtained from thestudy of genes alone. For example, proteins, not genes, areresponsible for the phenotypes of cells. It is impossible toelucidate mechanisms of disease, aging, and effects of the en-

vironment solely by studying the genome. Only through thestudy of proteins can protein modifications be characterizedand the targets of drugs identified.

Annotation of the genome. One of the first applications ofproteomics will be to identify the total number of genes in agiven genome. This functional annotation of a genome is

necessary because it is still difficult to predict genes accuratelyfrom genomic data (46). One problem is that the exon-intronstructure of most genes cannot be accurately predicted bybioinformatics (43). To achieve this goal, genomic information

will have to be integrated with data obtained from proteinstudies to confirm the existence of a particular gene.

Protein expression studies. In recent years, the analysis ofmRNA expression by various methods has become increasinglypopular. These methods include serial analysis of gene expres-sion (SAGE) (160) and DNA microarray technology (142,143). However, the analysis of mRNA is not a direct reflectionof the protein content in the cell. Consequently, many studieshave now shown a poor correlation between mRNA and pro-tein expression levels (2, 12, 67, 75). The formation of mRNA

FIG. 1. Types of proteomics and their applications to biology.

VOL. 66, 2002 MOLECULAR BIOLOGISTS GUIDE TO PROTEOMICS 41


4/25

is only the first step in a long sequence of events resulting in the

synthesis of a protein (Fig. 2). First, mRNA is subject to post-transcriptional control in the form of alternative splicing, poly-adenylation, and mRNA editing (117). Many different proteinisoforms can be generated from a single gene at this step.Second, mRNA then can be subject to regulation at the level ofprotein translation (78). Proteins, having been formed, aresubject to posttranslational modification. It is estimated thatup to 200 different types of posttranslational protein modifi-cation exist (89). Proteins can also be regulated by proteolysis(86) and compartmentalization (33). The average number ofprotein forms per gene was predicted to be one or two inbacteria, three in yeast, and three or more in humans (168).Therefore, it is clear that the tenet ofone gene, one protein

is an oversimplification. In addition, some bodily fluids such asserum or urine have no mRNA source and therefore cannot bestudied by mRNA analysis.

Protein function. According to one study, no function can beassigned to about one-third of the sequences in organisms for

which the genomes have been sequenced (47). The completeidentification of all proteins in a genome will aid the field ofstructural genomics in which the ultimate goal is to obtain 3-Dstructures for all proteins in a proteome. This is necessarybecause the functions of many proteins can only be inferred byexamination of their 3-D structure (24).

Protein modifications. One of the most important applica-tions of proteomics will be the characterization of posttrans-lational protein modifications. Proteins are known to be mod-ified posttranslationally in response to a variety of intracellularand extracellular signals (74). For example, protein phosphor-

ylation is an important signaling mechanism and disregulationof protein kinases or phosphatases can result in oncogenesis(74). By using a proteomics approach, changes in the modifi-cations of many proteins expressed by a cell can be analyzedsimultaneously.

Protein localization and compartmentalization. One of themost important regulatory mechanisms known is protein local-ization. The mislocalization of proteins is known to have pro-found effects on cellular function (e.g., cystic fibrosis) (42).Proteomics aims to identify the subcellular location of eachprotein. This information can be used to create a 3-D protein

map of the cell, providing novel information about protein

regulation.Protein-protein interactions. Of fundamental importance inbiology is the understanding of protein-protein interactions.The process of cell growth, programmed cell death, and thedecision to proceed through the cell cycle are all regulated bysignal transduction through protein complexes (127). Proteom-ics aims to develop a complete 3-D map of all protein inter-actions in the cell. One step toward this goal was recentlycompleted for the microorganism Helicobacter pylori (133).Using the yeast two-hybrid method to detect protein interac-tions, 1,200 connections were identified between H. pylori pro-teins covering 46.6% of the genome (133). A comprehensivetwo-hybrid analysis has also been performed on all the proteinsfrom the yeast S. cerevisiae (157).

Types of Proteomics

Protein expression proteomics. The quantitative study ofprotein expression between samples that differ by some vari-able is known as expression proteomics. In this approach, pro-tein expression of the entire proteome or of subproteomesbetween samples can be compared. Information from this ap-proach can identify novel proteins in signal transduction oridentify disease-specific proteins.

Structural proteomics. Proteomics studies whose goal is tomap out the structure of protein complexes or the proteins

present in a specific cellular organelle are known as cell mapor structural proteomics (21). Structural proteomics attemptsto identify all the proteins within a protein complex or or-ganelle, determine where they are located, and characterize allprotein-protein interactions. An example of structural pro-teomics was the recent analysis of the nuclear pore complex(137). Isolation of specific subcellular organelles or proteincomplexes by purification can greatly simplify the proteomicanalysis (83). This information will help piece together theoverall architecture of cells and explain how expression ofcertain proteins gives a cell its unique characteristics.

Functional proteomics. Functional proteomics is a broadterm for many specific, directed proteomics approaches. Insome cases, specific subproteomes are isolated by affinity chro-

FIG. 2. Mechanisms by which a single gene can give rise to multiple gene products. Multiple protein isoforms can be generated by RNAprocessing when RNA is alternatively spliced or edited to form mature mRNA. mRNA, in turn, can be regulated by stability and efficiency oftranslation. Proteins can be regulated by additional mechanisms, including posttranslational modification, proteolysis, or compartmentalization.



5/25

matography for further analysis. This could include the isola-tion of protein complexes or the use of protein ligands toisolate specific types of proteins. This approach allows a se-lected group of proteins to be studied and characterized andcan provide important information about protein signaling,disease mechanisms or protein-drug interactions.

TECHNOLOGY OF PROTEOMICS

An integral part of the growth of proteomics has been in theadvances made in protein technologies. Twenty-six years ago,

when 2-DE was introduced, very few tools existed for proteom-ics. Since that time, new technologies have emerged and oldones have been improved in areas from protein separation toprotein identification. However, it is also clear that it is still notfeasible to conduct many types of proteomics because of lim-itations in technology. These problems will have to be solvedand new technologies must be developed for proteomics toreach its full potential. A typical proteomics experiment (suchas protein expression profiling) can be broken down into the

following categories: (i) the separation and isolation of pro-teins from a cell line, tissue, or organism; (ii) the acquisition ofprotein structural information for the purposes of proteinidentification and characterization; and (iii) database utiliza-tion.

Separation and Isolation of Proteins

By the very definition of proteomics, it is inevitable thatcomplex protein mixtures will be encountered. Therefore,methods must exist to resolve these protein mixtures into theirindividual components so that the proteins can be visualized,identified, and characterized. The predominant technology for

protein separation and isolation is polyacrylamide gel electro-phoresis. Unlike the breakthroughs in molecular biology thateventually enabled the sequencing of the human genome,some aspects of protein science have shown little progress overthe years. Protein separation technology is one of them. Sinceits inception some 32 years ago (92), protein electrophoresisstill remains the most effective way to resolve a complex mix-ture of proteins. In many applications, it is at this stage wherethe bottleneck occurs. This is because 1- or 2-DE is a slow,tedious procedure that is not easily automated. However, untilsomething replaces this methodology, it will remain an essen-tial component of proteomics.

One- and two-dimensional gel electrophoresis. For many

proteomics applications, 1-DE is the method of choice to re-solve protein mixtures. In 1-DE, proteins are separated on thebasis of molecular mass. Because proteins are solubilized insodium dodecyl sulfate (SDS), protein solubility is rarely aproblem. Moreover, 1-DE is simple to perform, is reproduc-ible, and can be used to resolve proteins with molecular massesof 10 to 300 kDa. The most common application of 1-DE is thecharacterization of proteins after some form of protein purifi-cation. This is because of the limited resolving power of a 1-Dgel. If a more complex protein mixture such as a crude celllysate is encountered, then 2-DE can be used. In 2-DE, pro-teins are separated by two distinct properties. They are re-solved according to their net charge in the first dimension andaccording to their molecular mass in the second dimension.

The combination of these two techniques produces resolutionfar exceeding that obtained in 1-DE.

One of the greatest strengths of 2-DE is the ability to resolveproteins that have undergone some form of posttranslationalmodification. This resolution is possible in 2-DE because manytypes of protein modifications confer a difference in charge as

well as a change in mass on the protein. One such example isprotein phosphorylation. Frequently, the phosphorylated formof a protein can be resolved from the nonphosphorylated formby 2-DE. In this case, a single phosphoprotein will appear asmultiple spots on a 2-D gel (94). In addition, 2-DE can detectdifferent forms of proteins that arise from alternative mRNAsplicing or proteolytic processing.

The primary application of 2-DE continues to be proteinexpression profiling. In this approach, the protein expression ofany two samples can be qualitatively and quantitatively com-pared. The appearance or disappearance of spots can provideinformation about differential protein expression, while theintensity of those spots provides quantitative informationabout protein expression levels. Protein expression profiling

can be used for samples from whole organisms, cell lines,tissues, or bodily fluids. Examples of this technique include thecomparison of normal and diseased tissues (44) or of cellstreated with various drugs or stimuli (30, 57, 69, 141, 144). Anexample of 2-DE used in protein profiling is shown in Fig. 3.

Another application of 2-DE is in cell map proteomics.2-DE is used to map proteins from microorganisms (28, 146),cellular organelles (83), and protein complexes (134). It canalso be used to resolve and characterize proteins in subpro-teomes that have been created by some form of purification ofa proteome (26, 35, 38, 83). Because a single 2-DE gel canresolve thousands of proteins (30, 44, 146), it remains a pow-erful tool for the cataloging of proteins. Many 2-DE databases

have been constructed and are available on the World WideWeb (15).

A number of improvements have been made in 2-DE overthe years (13, 29). One of the biggest improvements was theintroduction of immobilized pH gradients, which greatly im-proved the reproducibility of 2-DE (20, 59). The use of fluo-rescent dyes has improved the sensitivity of protein detection(126), and specialized pH gradients are able to resolve moreproteins (59). The speed of running 2-DE has been improved,and 2-D gels can now be run in the minigel format (139). Inaddition, there have been efforts to automate 2-DE. Hoch-strassers group has automated the process of 2-DE from gelrunning to image analysis and spot picking (156). The use of

computers has aided the analysis of complex 2-D gel images(16). This is a critical aspect of 2-DE because a high degree ofaccuracy is required in spot detection and annotation if arti-facts are to be avoided. Recently, a molecular scanner wasdeveloped to record 2-DE images (19). Software programssuch as Melanie compare computer images of 2-D gels andfacilitate both the identification and quantitation of proteinspots between samples (171). A recent exciting advance in2-DE was developed by Minden and coworkers (158). Thistechnology is called difference gel electrophoresis (DIGE) andutilizes fluorescent tagging of two protein samples with twodifferent dyes. The tagged proteins are run on the same 2-Dgel, and postrun fluorescence imaging of the gel is used tocreate two images, which are superimposed to identify pattern



6/25

differences. The dyes are amine reactive and are designed toensure that proteins common to both samples have the samerelative mobility regardless of the dye used to tag them. Thistechnique circumvents the need to compare several 2-D gels.In their original paper, DIGE was used to detect differencesbetween exogenous proteins in two D. melanogaster embryoextracts at nanogram levels (158). Moreover, an inducible pro-tein from Escherichia coli was detected after 15 min of induc-tion. This technology is now commercially available from Am-

ersham/Pharmacia.However, a number of problems with 2-DE still remain.Despite efforts to automate protein analysis by 2-DE, it is stilla labor-intensive and time-consuming process. A typical 2-DEexperiment can take two days, and only a single sample can beanalyzed per gel. In addition, 2-DE is limited by both thenumber and type of proteins that can be resolved. For example,the protein mixture obtained from a eukaryotic cell lysate istoo complex to be completely resolved on a single 2-D gel (29).Many large or hydrophobic proteins will not enter the gelduring the first dimension, and proteins of extreme acidity orbasicity (proteins with pIs below pH 3 and above pH 10) arenot well represented (59). Some of these problems can beovercome with different solubilization conditions and pH gra-

dients (59). Another limitation of 2-DE is the inability todetect low-copy proteins when a total-cell lysate is analyzed(67, 96, 146). In a crude cell extract, the most abundant pro-teins can dominate the gel, making the detection of low-copyproteins difficult. It was determined in the analysis of yeastproteins by 2-DE that no proteins defined as low-copy proteins

were visible by 2-DE (67). Yet it is estimated that over half ofthe 6,000 genes in yeast may encode low-copy proteins (58). Inmammalian cells, the dynamic range of protein expression is

estimated to be between 7 and 9 orders of magnitude (36). Thisproblem cannot be overcome by simply loading more proteinon the gel, because the resolution will decrease and the comi-gration of proteins will increase (36). Because of these limita-tions, the largest application of 2-DE in the future will prob-ably involve the analysis of protein complexes or subproteomesas opposed to whole proteomes.

Alternatives to electrophoresis. The limitations of 2-DEhave inspired a number of approaches to bypass protein gelelectrophoresis. One approach is to convert an entire proteinmixture to peptides (usually by digestion with trypsin) and thenpurify the peptides before subjecting them to analysis by MS.Various methods for peptide purification have been devised,including liquid chromatography (95, 106, 174), capillary elec-

FIG. 3. Protein expression profiling by 2-DE. Whole-cell lysates from nontransformed and Abelson murine leukemia virus (AMuLV)-transformed mouse fibroblasts were resolved by 2-DE, and proteins were visualized by silver staining. Differentially expressed proteins were excisedfrom the gel and identified by MS.



7/25

trophoresis (55, 155), and a combination of techniques such asmultidimensional protein identification (95) or cation-ex-change chromatography and reverse-phase (RP) chromatogra-phy (120). The advantage of these methods is that because a2-D gel is avoided, a greater number of proteins in the mixturecan be represented. The disadvantage is that it can require an

immense amount of time and computing power to deconvolutethe data obtained. In addition, considerable time and effortmay be expended in the analysis of uninteresting proteins. Oneof the most exciting techniques to emerge as an alternative toprotein electrophoresis is that of isotope-coded affinity tags(ICAT). This method allows the quantitative protein profilingbetween different samples without the use of electrophoresis(see Proteomics applications below).

Acquisition of Protein Structure Information

Edman sequencing. One of the earliest methods used forprotein identification was microsequencing by Edman chemis-try to obtain N-terminal amino acid sequences. Little has

changed in Edman chemistry since its introduction, but im-provements in sequencing technology have increased the sen-sitivity and ease of Edman sequencing. Although the use ofEdman sequencing is waning in the field of proteomics, it isstill a very useful tool for several reasons. First, because Ed-man sequencing existed before MS as a sequencing tool, aconsiderable number of investigators continue to use Edmansequencing. Second, Edman sequencing of relatively abundantproteins is a viable alternative to MS if a mass spectrometer isin high demand for the identification of low-copy proteins or isnot available. Finally, Edman sequencing is used to obtain theN-terminal sequence of a protein (if possible) to determine itstrue start.

The N-terminal sequencing of proteins was introduced byEdman in 1949 (45). Today, Edman sequencing is most oftenused to identify proteins after they are transferred to mem-branes. The development of membranes compatible with se-quencing chemicals allowed Edman sequencing to become amore applicable sequencing method for the identification ofproteins separated by SDS-polyacrylamide gel electrophoresis(8, 159). One of the biggest problems that has limited thesuccess of Edman sequencing in the past is N-terminal modi-fication of proteins. Since it is difficult to tell if a protein isN-terminally blocked before it is sequenced, precious samples

were often lost in failed sequencing attempts. To overcome thisproblem, we developed a novel approach called mixed-peptide

sequencing (38). In mixed-peptide sequencing, a protein isconverted into peptides by cleavage with cyanogen bromide(CNBr) or skatole and the peptides are sequenced in an Ed-man sequencer simultaneously (9, 38, 99).

Briefly, the process of mixed-peptide sequencing involvesseparation of a complex protein mixture by polyacrylamide gelelectrophoresis (1-D or 2-D) and then transfer of the proteinsto an inert membrane by electroblotting (Fig. 4). The proteinsof interest are visualized on the membrane surface, excised,and fragmented chemically at methionine (by CNBr) or tryp-tophan (by skatole) into several large peptide fragments. Onaverage, three to five peptide fragments are generated, consis-tent with the frequency of occurrence of methionine and tryp-tophan in most proteins. The membrane piece is placed di-

rectly into an automated Edman sequencer without furthermanipulation. Between 6 and 12 automated Edman cycles arecarried out (4 to 8 h), and the mixed-sequence data are fed intothe FASTF or TFASTF algorithms, which sort and match thedata against protein (FASTF) and DNA (TFASTF) databasesto unambiguously identify the protein. The FASTF and

TFASTF programs were written in collaboration with WilliamPearson (Department of Biochemistry, University of Virginia).Because minimal sample handling is involved, mixed-peptidesequencing can be a sensitive approach for identifying proteinsin polyacrylamide gels at the 0.1- to 1-pmol level. An exampleof mixed-peptide sequencing is shown in Fig. 5A. The mixed-sequence approach has the advantage of enabling subsequentsearches to be carried out against unannotated or non-species-specific DNA databases as well as annotated protein databases.This is because the T/FASTF algorithms utilize actual aminoacid sequence and are therefore able to tolerate errors in thedatabase as well as polymorphisms or conservative substitu-tions. A recent variation of T/FASTF has been devised for MS(101) (Fig. 5B). The T/FASTF/S programs are available at

http://fasta.bioch.virginia.edu/ (Table 1).Mass spectrometry. MS enables protein structural informa-

tion, such as peptide masses or amino acid sequences, to beobtained. This information can be used to identify the proteinby searching nucleotide and protein databases (Fig. 4). It alsocan be used to determine the type and location of proteinmodifications. The harvesting of protein information by MScan be divided into three stages: (i) sample preparation, (ii)sample ionization, and (iii) mass analysis.

(i) Sample preparation. In most of proteomics, a protein isresolved from a mixture by using a 1- or 2-D polyacrylamidegel. The challenge is to extract the protein or its constituentpeptides from the gel, purify the sample, and analyze it by MS.

The extraction of whole proteins from gels is inefficient; how-ever, if a protein is in-gel digested with a protease, many ofthe peptides can be extracted from the gel. A method for in-gelprotein digestion was developed (149, 169) and is now com-monly applied to both 1- and 2-D gels (136). In-gel digestion ismore efficient at sample recovery than other common methodssuch as electroblotting (37). In addition, the conversion of aprotein into its constituent peptides provides more informationthan can be obtained from the whole protein itself. For manyapplications, the peptides recovered following in-gel digestionneed to be purified to remove gel contaminants. Commonimpurities from electrophoresis such as salts, buffers, and de-tergents can interfere with MS (172). In addition, peptide

samples often require concentration before being analyzed byMS. One method of peptide purification commonly employedfor this purpose is reverse-phase chromatography, which isavailable in a variety of formats. Peptides can be purified withZipTips (Millipore) or Poros R2 perfusion material (PerSep-tive Biosystems, Framingham, Mass.) (149, 169, 170) or byhigh-pressure liquid chromatography (HPLC).

(ii) Sample ionization. For biological samples to be analyzedby MS, the molecules must be charged and dry. This is accom-plished by converting them to desolvated ions. The two mostcommon methods for this are electrospray ionization (ESI)and matrix-assisted laser desorption/ionization (MALDI). Inboth methods, peptides are converted to ions by the additionor loss of one or more protons. ESI and MALDI are soft



8/25

ionization methods that allow the formation of ions withoutsignificant loss of sample integrity. This is important because it

enables accurate mass information to be obtained about pro-teins and peptides in their native states.

(a) Electrospray ionization. In ESI, a liquid sample flowsfrom a microcapillary tube into the orifice of the mass spec-trometer, where a potential difference between the capillaryand the inlet to the mass spectrometer results in the generationof a fine mist of charged droplets (52, 72, 172). As the solventevaporates, the sizes of the droplets decrease, resulting in theformation of desolvated ions (52). A significant improvementin ESI technology occurred with the development of nanosprayionization (169, 170). In nanospray ionization, the microcapil-lary tube has a spraying orifice of 1 to 2 m and flow rates aslow as 5 to 10 nl/min (170). The low flow rates possible with

nanospray ionization reduce the amount of sample consumedand increase the time available for analysis (148, 149). For ESI,there are several ways to deliver the sample to the mass spec-trometer. The simplest method is to load individual microcap-illary tubes with sample. Because a new microcapillary tube isused for each sample, cross-contamination is avoided. In ESI,peptides require some form of purification after in-gel diges-tion, and this can be accomplished directly in the microcapil-lary tubes. The drawback to both the purification and manualloading of microcapillary tubes is that it is tedious and slow. Asan alternative, electrospray sources have been connected inline with liquid chromatography (LC) systems that automati-cally purify and deliver the sample to the mass spectrometer.Examples of this method are LC (39, 55, 95, 106), reverse-

phase LC (RP-LC) (64) and reverse-phase microcapillary LC(RP-LC) (41).

(b) Matrix-assisted laser desorption/ionization. In MALDI, thesample is incorporated into matrix molecules and then sub-

jected to irradiation by a laser. The laser promotes the forma-tion of molecular ions (84). The matrix is typically a smallenergy-absorbing molecule such as 2,5-dihydroxybenzoic acidor -cyano-4-hydroxycinnamic acid. The analyte is spotted,along with the matrix, on a metal plate and allowed to evapo-rate, resulting in the formation of crystals. The plate, which canbe 96-well format, is then placed in the mass spectrometer, andthe laser is automatically targeted to specific places on theplate. Since sample application can be performed by a robot,the entire process including data collection and analysis can beautomated. This is the single biggest advantage of MALDI.

Another advantage of MALDI over ESI is that samples canoften be used directly without any purification after in-geldigestion (131).

(iii) Mass analysis. Mass analysis follows the conversion ofproteins or peptides to molecular ions. This is accomplished bythe mass analyzers in a mass spectrometer, which resolve themolecular ions on the basis of their mass and charge in a

vacuum.(a) Quadrupole mass analyzers. One of the most common

mass analyzers is the quadrupole mass analyzer. Here, ions aretransmitted through an electric field created by an array of fourparallel metal rods, the quadrupole (172). A quadrupole canact to transmit all ions or as a mass filter to allow the trans-mission of ions of a certain mass-to-charge (m/z) ratio. If mul-

FIG. 4. Strategies for protein identification. The identification of proteins from a polyacrylamide gel by mixed-peptide sequencing or MS isdepicted. For mixed-peptide sequencing, proteins are transferred to a membrane and cleaved with CNB or skatole, and the resulting peptides aresequenced simultaneously by Edman degradation. For MS, proteins are in-gel digested with proteases and the resulting peptides are massfingerprinted or sequenced. Information from all these methods is used to search nucleotide and protein databases for protein identification.



9/25

tiple quadrupoles are combined, they can be used to obtaininformation about the amino acid sequence of a peptide. For amore detailed review of the operating principles of a quadru-pole mass analyzer, the reader is directed to several excellentreviews (25, 109, 172).

(b) Time of flight. A time-of-flight (TOF) instrument is oneof the simplest mass analyzers. It measures the m/z ratio of anion by determining the time required for it to traverse thelength of a flight tube. Some TOF mass analyzers include anion mirror at the end of the flight tube, which reflects ions back

through the flight tube to a detector. In this way, the ion mirrorserves to increase the length of the flight tube. The ion mirroralso corrects for small energy differences among ions (172).Both of these factors contribute to an increase in mass reso-lution.

(c) Ion trap. Ion trap mass analyzers function to trap molec-ular ions in a 3-D electric field. In contrast to a quadrupolemass analyzer, in which ions are discarded before the analysisbegins, the main advantage of an ion trap mass analyzer is theability to allow ions to be stored and then selectively ejected

FIG. 5. The FASTF and FASTS search programs. (A) Example of a FASTF search where the amino acid sequence is obtained by Edmansequencing of a mixture of peptides. The information is then deconvoluted by a computer algorithm, and the results are given an expectation score

(e). (B) With the FASTS program, a similar type of search is conducted except that peptide sequences obtained from MS are used.



10/25

from the ion trap, increasing sensitivity (172). For a review ofthe operating principles of an ion trap mass spectrometer, seereference 34.

(iv) Types of mass spectrometers. Most mass spectrometers

consist of four basic elements: (i) an ionization source, (ii) oneor more mass analyzers, (iii) an ion mirror, and (iv) a detector.The names of the various instruments are derived from thename of their ionization source and the mass analyzer. Some ofthe most common mass spectrometers are discussed; for amore comprehensive review of mass spectrometers, the readeris directed to references (76 and 172). The analysis of proteinsor peptides by MS can be divided into two general categories:(i) peptide mass analysis and (ii) amino acid sequencing. Inpeptide mass analysis or peptide mass fingerprinting, themasses of individual peptides in a mixture are measured andused to create a mass spectrum (70). In amino acid sequencing,a procedure known as tandem mass spectrometry, or MS/MS,

is used to fragment a specific peptide into smaller peptides,which can then be used to deduce the amino acid sequence.

(a) Triple quadrupole. Triple-quadrupole mass spectrome-ters are most commonly used to obtain amino acid sequences.In the first stage of analysis, the machine is operated in MSscan mode and all ions above a certain m/z ratio are transmit-ted to the third quadrupole for mass analysis (Fig. 6) (82, 173).In the second stage, the mass spectrometer is operated inMS/MS mode and a particular peptide ion is selectively passedinto the collision chamber. Inside the collision chamber, pep-tide ions are fragmented by interactions with an inert gas by aprocess known as collision-induced dissociation or collisionallyactivated dissociation. The peptide ion fragments are then

resolved on the basis of their m/z ratio by the third quadrupole(Fig. 6). Since two different mass spectra are obtained in thisanalysis, it is referred to as tandem mass spectrometry (MS/MS). MS/MS is used to obtain the amino acid sequence ofpeptides by generating a series of peptides that differ in massby a single amino acid (71, 73).

(b) Quadrupole-TOF. In recent years, several hybrid massspectrometers have emerged from the combination of differentionization sources with mass analyzers. One example is thequadrupole-TOF mass spectrometer (111, 112, 162). In thismachine, the first quadrupole (Q1) and the quadrupole colli-sion cell (q) of a triple-quadrupole machine have been com-bined with a time-of-flight analyzer (TOF) (145). The mainapplications of a QqTOF mass spectrometer are protein iden-

tification by amino acid sequencing and characterization ofprotein modifications. However, because it is coupled to elec-trospray, it is not typically utilized for large-scale proteomics.

(c) MALDI-TOF. The principal application of a MALDI-

TOF mass spectrometer is peptide mass fingerprinting becauseit can be completely automated, making it the method ofchoice for large-scale proteomics work (48). Because of itsspeed, MALDI-TOF is frequently used as a first-pass instru-ment for protein identification. If proteins cannot be identifiedby fingerprinting, they can then be analyzed by electrosprayand MS/MS. A MALDI-TOF machine can also be used toobtain the amino acid sequence of peptides by a methodknown as post-source decay (152). However, peptide sequenc-ing by post-source decay is not as reliable as sequencing withcompeting electrospray methods because the peptide fragmen-tation patterns are much less predictable (85, 111).

(d) MALDI-QqTOF. The MALDI-QqTOF mass spectrom-

eter was developed to permit both peptide mass fingerprintingand amino acid sequencing (97, 147). It was formed by thecombination of a MALDI ion source with a QqTOF massanalyzer (63, 91, 97, 147, 162). Thus, if a sample is not identi-fied by peptide mass fingerprinting in the first step, the aminoacid sequence can then be obtained without having to use adifferent mass spectrometer. However, in our experience, theamino acid sequence information obtained using this instru-ment was more difficult to interpret than that obtained from ananospray-QqTOF mass spectrometer.

(e) FT-ICR. A Fourier transform ion cyclotron resonance(FT-ICR) mass spectrometer is an ion-trapping instrumentthat can achieve higher mass resolution and mass accuracy

than any other type of mass spectrometer (10). Recently, FT-ICR has been employed in the analysis of biomolecules ionizedby both ESI and MALDI. The unique abilities of FT-ICRprovide certain advantages compared to other mass spectrom-eters. For example, because of its high resolution, FT-ICR canbe used for the analysis of complex mixtures. FT-ICR, coupledto ESI, is also being employed in the study of protein interac-tions and protein conformations. A high-throughput, large-scale proteomics approach involving FT-ICR has recently beendeveloped by Smith et al. (150). For a review of the operatingprinciples of FT-ICR and its applications, the reader is di-rected to reference 104.

(v) Peptide fragmentation. As peptide ions are introducedinto the collision chamber, they interact with the collision gas

TABLE 1. World Wide Web tools for searching databases with protein information obtained either from mass spectrometryor from Edman degradation

Site name URL Information available Reference

MOWSE http://srs.hgmp.mrc.ac.uk/cgi-bin/mowse Peptide mass mapping and sequencing 125ProFound http://prowl.rockefeller.edu/cgi-bin/ProFound Peptide mass mapping and sequencing 176PeptIdent http://www.expasy.ch/tools/peptident. Peptide mass mapping and sequencing 165

PepSea http://195.41.108.38/PepSeaIntro.html Peptide mass mapping and sequencing 102MASCOT http://www.matrixscience.com/ Peptide mass mapping and sequencing 129PepFrag http://www.proteometrics.com/ Peptide mass mapping and sequencing 54Protein Prospector http://prospector.ucsf.edu/ Peptide mass mapping and sequencing 32FindMod http://www.expasy.ch/tools/ findmod/ Posttranslational modification 166SEAQUEST http:// fields.scripps.edu/sequest/ Uninterpreted MS/MS searching 49FASTA Search Programs http://fasta.bioch.virginia.edu/ Protein and nucleotide database searching 101Cleaved Radioactivity of

Phosphopeptideshttp://fasta.bioch.virginia.edu/crp Protein phosphorylation site mapping MacDonald et al.

submitted



11/25

(usually nitrogen or argon) and undergo fragmentation pri-marily along the peptide backbone (71, 73, 172). Since peptidescan undergo multiple types of fragmentation, nomenclaturehas been created to indicate what type of ions have beengenerated (Fig. 7). If, after peptide bond cleavage, the chargeis maintained on the N-terminus of the ion, it is designated ab-ion, whereas if the charge is maintained on the C terminus,

it is a y-ion (Fig. 7) (18, 135, 173). The difference in massbetween adjacent y- or b-ions corresponds to that of an aminoacid. This can be used to identify the amino acid and hence thepeptide sequence, with the exception of isoleucine and leucine,

which are identical in mass and therefore indistinguishable(103). Both y- and b-type ions can also eliminate NH3 (17Da), H2O (18 Da) and CO (28 Da), resulting in pairs ofsignals observed in the mass spectrum (Fig. 7). In addition tofragmentation along the peptide backbone, cleavage can occuralong amino acid side chains, and this information can be usedto distinguish isoleucine and leucine (172).

(vi) Our approach to mass spectrometry. The sensitivity of amass spectrometer is probably the single most important fea-

ture of the instrument. What is the sensitivity of a modern massspectrometer? How much protein is needed to make an un-ambiguous identification? Many factors can affect sensitivity,such as sample preparation, sample ionization, the type ofmass spectrometer used, the sample itself, and the type ofdatabase search employed. In our laboratory, we rely on 1- or2-DE electrophoresis for the isolation and visualization of pro-tein targets. We typically stain our gels with either Coomassieblue or silver stain. For most proteins, staining with Coomassieblue will give a dark band for 1g of protein and a discern-ible one for 200 ng. With silver staining, we can detect a darkband at 50 ng and faint yet discernible bands at 5 to 10 ng.However, a significant number of proteins do not stain well bythese methods and larger proteins tend to bind more stain

(mole/mole) than small proteins. In addition, MS is not aquantitative technique because peptide ionization is not quan-titative. Therefore, some proteins that are barely visible on gelscan give stronger signals by MS than do some darkly stainingproteins. For example, one of the most frequently sequencedproteins in MS is human keratin, a component of dust. It is acontaminant that will often appear on polyacrylamide gels as

faint silver-stained bands with a variety of molecular weights. Itcan be introduced simply from the glass plates or gel combsused for protein gels; therefore, it is a good idea to wash theseitems in concentrated acid before use.

We have found in our laboratory that most proteins appliedto the gel at 5 to 10 ng (100 to 200 fmol for a 50-kDa protein)can be identified by MS. However, the ability to identify aprotein depends on the protein itself and its presence in thedatabase. Below 5 to 10 ng, the success rate decreases becausefewer peptides are obtained for sequencing. Several prominentMS laboratories routinely report record-breaking sequencingsensitivity to the attomolar level. However, this sensitivity is

FIG. 6. MS/MS. Conventional and MS/MS modes of analysis in a triple-quadrupole mass spectrometer are shown. (A) In the normal scanningmode, all ions of a certain m/z range are transmitted through the first two quadrupoles for mass analysis in the third quadrupole. From this MSspectrum, a parent ion is selected for fragmentation in the collision cell. (B) In MS/MS mode, the parent ion is selectively transmitted into thecollision chamber and fragmented, and the resulting daughter ions are resolved in the third quadrupole.

FIG. 7. Peptide ion fragmentation nomenclature. Low-energy col-lisions promote fragmentation of a peptide primarily along the peptidebackbone (73). Peptide fragmentation which maintains the charge onthe C terminus is designated a y-ion, whereas fragmentation whichmaintains the charge on the N terminus is designated a b-ion. Addi-tional types of fragmentation are also indicated.



12/25

usually toward a purified peptide sample that is directly intro-duced into the mass spectrometer. Since most proteins areisolated from gels for identification, this is not an accuratemeasure of sensitivity. In another case, it was reported that anamino acid sequence was obtained after the in-gel digestion of25 fmol (1.7 ng) of pure bovine serum albumin (90). Again,

since the protein was known before the analysis began, this isnot a fair assessment of sensitivity. For unknown proteins,more protein is required because several peptides have to besequenced before a confident assignment can be made.

A typical approach to protein identification in our laboratoryis outlined in Fig. 8. Protein from a polyacrylamide gel isexcised and then in-gel digested with trypsin by the method ofWilm et al. (170). Following peptide extraction from the gel,

we purify the peptides on Poros R2 (149, 169) in microcapillarytubes by using the method described on the website http://www.protana.com/products/applicationnotes/puri fication/default.asp. We use the API QSTAR Pulsar mass spectrometer (AB/MDS-SCIEX) with nanospray ionization to obtain an MS scanof the peptide mixture. From the MS scan, a peptide ion is

selected for MS/MS based on its signal strength and chargestate, which allow it to be distinguished from the backgroundions. In nanospray ionization, most peptide ions are eitherdoubly or triply charged whereas the background ions aresingly charged. This peptide ion is also known as the parention. MS/MS of a parent ion is performed, and amino acidsequence information for the peptide is obtained. As shown inFig. 8, a single peptide was sequenced and found to matchrhoptry-associated protein 2 (RAP-2) from Plasmodium falci-parum. Since matching multiple peptides to a protein increasesthe confidence of identification (106), we typically sequenceseveral peptides for each sample. For RAP-2, a total of fourpeptides were found to match the protein. Because the staining

intensity on gels is not always a good indicator of the signalobtained by MS and because gel bands often contain proteinmixtures, additional criteria can aid in protein identification.For example, if the major protein excised from the gel was 50kDa, does the protein identified match in molecular mass? Isthe protein from the expected species? If a protein is isolatedfrom a 2-D gel, does it match the expected isoelectric point asexhibited on the gel?

Database Utilization

Databases allow protein structural information harvestedfrom Edman sequencing or MS to be used for protein identi-

fication. The goal of database searching is to be able to quicklyand accurately identify large numbers of proteins (132). Thesuccess of database searching depends on the quality of thedata obtained in the mass spectrometer, the quality of thedatabase searched, and the method used to search the data-base. What is the best way to identify an unknown protein?What type of database search engine should be used?

Peptide mass fingerprinting database searching. Onemethod of protein identification is peptide mass fingerprinting(77, 79, 102, 125, 175). In this method, the masses of peptidesobtained from the proteolytic digestion of an unknown proteinare compared to the predicted masses of peptides from thetheoretical digestion of proteins in a database (Fig. 9). Ifenough peptides from the real mass spectrum and the theo-

retical one overlap, a protein identification can be made. Theprincipal advantage of peptide mass fingerprinting is speed.The analysis and database search can be fully automated.

The single biggest disadvantage of peptide mass fingerprint-ing is ambiguity in protein identification. This is because ofpeptide mass redundancy. For example, a peptide of 5 amino

acids can have the same mass by simple rearrangement of itsconstitutive amino acids; e.g., peptide VAGSE has the samemass as AVGSE or AEVGS and so on. For this technique tobe successful, the masses of a large number of peptides must beobtained to provide enough specificity in the search, and this isnot always possible. Mass redundancy occurs with greater fre-quency in large genomes. Moreover, peptide mass fingerprint-ing is effective only in the analysis of proteins from organisms

whose genome is small, completely sequenced, and well anno-tated (131). It has limited use against unannotated or untrans-lated DNA databases such as the human genome. Becausemass fingerprinting is not error tolerant, several factors inaddition to mass redundancy contribute to its limited use,including sequencing errors, conservative substitutions, poly-

morphisms, and six possible translations at the DNA level.Another factor affecting the success of peptide mass finger-

printing is mass accuracy (32, 62). Because it is critical toobtain an accurate measurement of the masses of multiplepeptides, factors that alter the masses of those peptides canreduce the success of the method. One such example is theposttranslational modification of proteins. If the unknown pro-tein is extensively modified, the peptides produced from thatprotein will not match the unmodified protein in the database.Recent improvements in the mass accuracy of mass spectrom-eters has increased the success rate of protein identification bythis method (32, 54).

Finally, peptide mass fingerprinting does not work well with

protein mixtures. As a protein mixture is converted to a mix-ture of peptides, it increases the complexity of the peptidemass fingerprint. The process of protein identification can behindered if even two or three proteins are present in thesample (107). Several search methods have emerged to accom-modate peptide mixtures in the mass spectrum. One example isa program called ProFound, which enables protein identifica-tion in simple protein mixtures (176). However, the lack ofability to analyze protein mixtures remains a major limitationof this method. A variety of tools for database searching nowexist on the World Wide Web (Table 1). The ExPASy serverprovides a variety of tools for proteomics and programs forprotein identification (reviewed in reference 165). Search pro-

grams used for peptide mass fingerprinting include PepSea(102), PeptIdent/MultiIdent (165), MS-Fit (32), MOWSE(125), and ProFound (176).

Amino acid sequence database searching. The most specifictype of database searching for protein identification uses pep-tide amino acid sequence. If the amino acid sequence of apeptide can be identified, it can be used to search databases tofind the protein from which it was derived. One method whichutilizes this information is peptide mass tag searching. In thismethod, a partial amino acid sequence is obtained by interpre-tation of the MS/MS spectrum (the sequence tag) and thisinformation is combined with the mass of the peptide and themasses of the peptide on either side of the sequence tag wherethe sequence is not known (Fig. 10). Also included in the



13/25

search is the type of protease used to produce the peptides.Peptide mass tag searching is a more specific tool for proteinidentification than peptide mass fingerprinting (49, 103, 115,170). In addition, one of the biggest advantages of utilizing

MS/MS to obtain peptide amino acid sequence is that, unlikepeptide mass fingerprinting, it is compatible with protein mix-tures. The ability to identify proteins in mixtures is one of thegreat advantages of using MS as a protein identification tool.

FIG. 8. Protein identification by MS/MS. (A) Protein from P. falciparum was resolved on a one-dimensional polyacrylamide gel, excised, andin-gel digested with trypsin. The resulting peptides were ionized by electrospray and analyzed by a Quadrupole-TOF mass spectrometer. (B) TheMS spectrum produced was scanned, and a parent ion of 678.8 was selected for fragmentation. (C) Enlargement of the parent ion peak at 678shown in panel B. The multiplet of peaks is due to the contribution in mass from the naturally occurring isotope 13C. A mass difference betweenthe peaks of 0.5 Da indicates that the peptide is doubly charged. (D) MS/MS scan of the 678 parent ion and analysis of the daughter ions produced.All y-ions (except for y-11) produced from fragmentation of the peptide are shown. (E) Identification of rhoptry-associated protein-2 usingBioAnalyst software (Applied Biosystems, Foster City, Calif.).



14/25

For example, in our laboratory we frequently identify multipleproteins from what appears to be a single band on an SDS-gel.In fact, in the majority of proteomics experiments, proteins arepresent in mixtures at the time of analysis.

The major disadvantage of performing MS/MS is that theprocess is not easily automated. As a result, considerable timeis expended in performing the analysis and interpreting themass spectrum. Although computer programs can assist in the

interpretation of the spectrum, they currently are not able tomake accurate assignments without some guidance. In addi-tion, when searching a database with peptide mass tags, thereis a lack offlexibility in the search programs. If a single mistakeis made in the assignment of a y- or b-ion (which can happenquite frequently), the amino acid sequence will be incorrectand the database search will bring up irrelevant proteins. Oftenit is necessary to confirm that the peptide sequence obtainedfrom the database matches the sequence obtained in the massspectrometer. This can be done by performing a theoreticalfragmentation of the peptide from the database and comparing

the two mass spectra. Additional clues can also be used, suchas verifying if the peptide obtained from the database ends inamino acids consistent with the type of protease used.

De novo peptide sequence information. Another approachto protein identification is to obtain de novo sequence datafrom peptides by MS/MS and then use all the peptide se-quences to search appropriate databases. Multiple peptide se-quences can be used for protein identification by searching

databases with the FASTS program (Mackey et al., submitted)(Fig. 5). The single biggest advantage of this method is thecapability of searching peptide sequence information acrossboth DNA and protein databases. This is because the searchengine utilized exhibits a certain amount of flexibility in theassignment of protein scores. This search method is useful fororganisms that do not have well-annotated databases such asXenopus laevis or human. However, because this method re-quires several peptide amino acid sequences of 3 or 4 aminoacids, it is not the first choice for peptide identification. Rather,the much faster methods of peptide mass fingerprinting orpeptide mass tag searching can be used first. If these searchmethods fail, de novo sequence information can be obtained

and used to identify the protein.Uninterpreted MS/MS data searching. A large number ofprograms are now available for the identification of proteins byusing uninterpreted MS/MS data. Examples include programssuch as Mascot (129), SONAR (53), and SEQUEST (49) (Ta-ble 1). However, searches against unannotated or untranslatedDNA databases with uninterpreted MS/MS data are likely tosuffer from the same pitfalls associated with mass fingerprint-ing. In particular, polymorphisms, sequencing errors, and con-servative substitutions will probably contribute to failure toaccurately identify a protein. The development of uninter-preted MS/MS search algorithms that are error tolerant mayovercome some of these shortcomings, provided that they as-sign some form of statistical scoring to the identified proteins.

FIG. 9. Strategy of protein identification by peptide mass fingerprinting. (A) The unknown protein is excised from a gel and converted topeptides by the action of a specific protease. The mass of the peptides produced is then measured in a mass spectrometer. (B) The mass spectrumof the unknown protein is searched against theoretical mass spectra produced by computer-generated cleavage of proteins in the database.

FIG. 10. Peptide mass tag searching. Shown is a schematic of howinformation from an unknown peptide (top) is matched to a peptidesequence in a database (bottom) for protein identification. The partialamino acid sequence or tag obtained by MS/MS is combined with thepeptide mass (parent mass), the mass of the peptide at the start of thesequence (mass tag 1), and the mass of the peptide at the end of thesequence (mass tag 2). The specificity of the protease used (trypsin isshown) can also be included in the search (103).



15/25

PROTEOMICS APPLICATIONS

The single most common application of proteomics is pro-tein identification. Most investigators use proteomics ap-proaches to isolate and display proteins based on their ownspecific criteria and then identify the proteins. Protein identi-fication provides immediate information that will direct subse-quent experimentation. For example, the identity of a proteincan reveal an expected result, validate a proteomics approach,provide completely unexpected information, or reveal that

your biochemical method is not working at all. We feel that themost critical stage of any proteomics approach is the strategicdesign for the isolation of protein targets. In recent years, asthe technology of MS has improved, there has been a de-emphasis on the front-end of proteomics experiments com-pared to data analysis. This can result in the isolation of hun-dreds of irrelevant proteins for identification, consuming bothtime and effort. Our general strategy is to devise techniquesthat enrich for low-abundance proteins and then analyze onlythe proteins that appear on differential display or are isolated

by affinity chromatography. To accomplish this, we use affinitycolumns and other strategies to select for protein targets. Ineach case, protein samples are subjected to a series of precol-umns and high-stringency washes to remove nonspecific pro-teins. This reduces the number of irrelevant proteins for anal-

ysis.

Characterization of Protein Complexes

Many laboratories are now engaged in an effort to charac-terize protein complexes by MS. Examples include Link et al.utilizing multidimensional LC and MS/MS to identify proteins(95) or Mann and colleagues identifying proteins present after

immunoprecipitation of protein complexes (124). Recently,Macara, Haystead, and coworkers used MS to identify inter-acting proteins with the Cdc42 effector, Borg3 (80). In thiscase, the bait protein, Borg3, was produced as a glutathioneS-transferase (GST) fusion in E. coli and then mixed with NIH3T3 cell lysate. Four interacting proteins were identified bymixed-peptide sequencing: heat shock protein Hsp70 andthree septins including Septin6, Cdc10, and Nedd5 (Fig. 11).None of these proteins were present in the GST-only controlsample. Although the interaction with Hsp70 was not pursued,it was shown from coimmunoprecipitation studies that endog-enous Borg3 interacts with endogenous Cdc10 and Nedd5 (80).

Additional proof from expression and structure-function stud-

ies confirmed a role for the Borg proteins as regulators ofseptin organization. It should be noted that although severalproteins were quickly identified as Borg3 interactors by thepull-down experiment, it took several more months of work toconfirm this interaction.

Protein Expression Profiling

The largest application of proteomics continues to be pro-tein expression profiling. Through the use of two-dimensionalgels or novel techniques such as ICAT, the expression levels ofproteins or changes in their level of modification between twodifferent samples can be compared and the proteins can be

identified. This approach can facilitate the dissection of signal-ing mechanisms or identify disease-specific proteins.

Expression profiling by two-dimensional electrophoresis.

Currently, the majority of protein expression profiling studiesare performed by 2-DE. Several diseases have been studied,including heart disease (44) and cancer (30). Cancer cells aregood candidates for proteomics studies because they can becompared to their nontransformed counterparts. Analysis ofdifferentially expressed proteins in normal versus cancer cells

can (i) identify novel tumor cell biomarkers that can be usedfor diagnosis, (ii) provide clues to mechanisms of cancer de-

velopment, and (iii) identify novel targets for therapeutic in-tervention. Protein expression profiling has been used in thestudy of breast (121), esophageal (121), bladder (30) and pros-tate (114) cancer. From these studies, tumor-specific proteins

were identified and 2-D protein expression databases weregenerated. Many of these 2-D protein databases are now avail-able on the World Wide Web (15).

Isotope-coded affinity tags. Recently, a novel method forprotein expression profiling was introduced that does not de-pend on the separation of proteins by 2-DE. This method isknown as isotope-coded affinity tags (ICAT) and relies on the

labeling of protein samples from two different sources with twochemically identical reagents that differ only in mass as a resultof isotope composition (66). Differential labeling of samples bymass allows the relative amount of protein between two sam-ples to be quantitated in the mass spectrometer. An example ofthe methodology of ICAT is shown in Fig. 12. Cell extract fromtwo different samples is reacted with one of two forms of theICAT reagent, an isotopically light form in which the linkercontains eight hydrogens or a heavy form in which the linkercontains eight deuterium atoms. The ICAT reagent reacts withcysteine residues in proteins via a thiol-reactive group andcontains a biotin moiety to facilitate purification (Fig. 12).Peptides are recovered on the basis of the biotin tag by avidinaffinity chromatography and are then analyzed by MS. The

FIG. 11. Identification of novel protein interactions by protein co-precipitation. (A) Pull-down experiment with a control (GST) or target(GST-Borg3) protein using 35S-labeled NIH 3T3 cell lysate. (B) Large-scale affinity purification of GST-Borg3 from the NIH 3T3 lysate.Individual proteins were microsequenced by mixed-peptide sequenc-

ing and identified by database searching with the FASTF algorithm(101).



16/25

difference in peak heights between heavy and light peptide ionsdirectly correlates with the difference in protein abundance inthe cells. Thus, if a protein is present at a threefold higher levelin one sample, this will be reflected in a threefold difference inpeak heights. Following quantitation of the peptides, they canbe fragmented by MS/MS and the amino acid sequence can beobtained. Thus, using this approach, proteins can be identifiedand their expression levels can be compared in the same anal-

ysis.The single biggest advantage of this method is the elimina-

tion of the 2-D gel for protein quantitation. As a result, anincreased amount of sample can be used to enrich for low-abundance proteins. Alternatively, the cell lysate can be frac-

tionated prior to reaction with the ICAT reagent. This canallow the enrichment of low-abundance proteins before theanalysis begins. The main disadvantages are that currently thismethod works only for proteins containing cysteine, eventhough this includes the majority of proteins (68). In addition,peptides must contain appropriately spaced protease cleavagesites flanking the cysteine residues. Finally, the ICAT label islarge (500 kDa) and remains with each peptide throughoutthe analysis. This can make database searching more difficult,especially for small peptides with limited sequence (4, 65).Sensitivity may also be of concern since tagged peptides de-rived from low-copy proteins are likely to be poorly recoveredduring the affinity step as a result of nonspecific interactions

FIG. 12. The ICAT method for measuring differential protein expression. (A) Structure of the ICAT reagent. ICAT consists of a biotin affinitygroup, a linker region that can incorporate heavy (deuterium) or light (hydrogen) atoms, and a thiol-reactive end group for linkage to cysteines.(B) ICAT strategy. Proteins are harvested from two different cell states and labeled on cysteine residues with either the light or heavy form of theICAT reagent. Following labeling, the two protein samples are mixed and digested with a protease such as trypsin. Peptides labeled with the ICATreagent can be purified by virtue of the biotin tag by using avidin chromatography. Following puri fication, ICAT-labeled peptides can be analyzedby MS to quantitate the peak ratios and proteins can be identified by sequencing the peptides with MS/MS.



17/25

with avidin-Sepharose. Studies have been performed to opti-mize the labeling of proteins with the ICAT reagent (151).

Protein arrays. Protein arrays are undergoing rapid devel-opment for the detection of protein-protein interactions andprotein expression profiling (17, 98, 180, 181). Recently, pro-tein microarrays were created using ordinary laboratory equip-

ment (98). Proteins were immobilized by being covalently at-tached to glass microscope slides, and the protein microarrayswere shown to be capable of interacting with other proteins,small molecules, and enzyme substrates (98). In another re-port, 5,800 yeast proteins were expressed and printed ontomicroscope slides. These protein microarrays were used toidentify novel calmodulin- and phospholipid-interacting pro-teins (180). These reports indicate that protein arrays holdgreat promise for the global analysis of protein-protein andprotein-ligand interactions. Undoubtedly, these arrays will im-prove as the technology for their creation is developed andrefined.

Proteomics Approach to Protein Phosphorylation

Posttranslational modification of proteins is a fundamentalregulatory mechanism, and characterization of protein modi-fications is paramount for understanding protein function. MSis one of the most powerful tools for the analysis of proteinmodifications because virtually any type of protein modifica-tion can be identified. Although we focus here on proteinphosphorylation, the analysis of other types of protein modi-fication by MS has been described (25). Protein phosphoryla-tion is one of the most common of all protein modi ficationsand has been found in nearly all cellular processes (74, 88,153). MS can be used to identify novel phosphoproteins, mea-sure changes in the phosphorylation state of proteins in re-

sponse to an effector, and determine phosphorylation sites inproteins. Identification of phosphorylation sites can provideinformation about the mechanism of enzyme regulation andthe protein kinases and phosphatases involved. A proteomicsapproach to protein phosphorylation has the advantage thatinstead of studying changes in the phosphorylation of a singleprotein in response to some perturbation, one can study all thephosphoproteins in a cell (the phosphoproteome) at the sametime. A common approach to studying protein phosphoryla-tion events is the use of in vivo labeling of phosphoproteins

with inorganic 32P. The phosphoproteomes of cells that differin some way (e.g., normal versus diseased) can be analyzed bygrowing cells in inorganic 32P and creating cell lysates. Changes

in the phosphorylation state of proteins can then be examinedby 2-DE and autoradiography. Proteins of interest are excisedfrom the gel and microsequenced by MS. A major limitation ofthis approach is that while many phosphorylated proteins canbe visualized by autoradiography, they cannot be identifiedbecause of their low abundance. One solution to this problemis enrichment of the phosphoproteome.

Phosphoprotein enrichment. Enrichment of the phospho-proteome of a cell can allow the identification of low-copyphosphoproteins that would otherwise go undetected. In oneapproach, phosphoproteins were enriched by conversion ofphosphoserine residues to biotinylated residues (118). Thismethod is an extension of techniques originally developed byHielmeyer and colleagues (108) and more recently by our

laboratory (51) for the identification of phosphorylation sitesusing Edman sequencing. Following derivatization, proteinsthat were formerly phosphorylated can be isolated by avidinaffinity chromatography (118). Proteins immobilized on avidinbeads can then be eluted with biotin, theoretically resulting inthe isolation of the entire phosphoserine proteome (Fig. 13).

By increasing the amount of cell lysate used for avidin affinitychromatography, low-abundance phosphoproteins can be en-riched. However, this technique does not work for phosphoty-rosine and the reactivity of phosphothreonine by this method is

very poor (118). Tyrosine-phosphorylated proteins can be iso-lated by the use of antiphosphotyrosine antibodies (124). As analternative, another method for phosphopeptide enrichment

was devised to allow the recovery of proteins phosphorylatedon serine, threonine, and tyrosine (179). In this method, aprotein or mixture of proteins is digested to peptides with aprotease and then subjected to a multistep procedure for theconversion of phosphoamino acids into free sulfhydryl groups.To capture the derivatized peptides, the free sulfhydryl groupsin the peptides are then reacted with iodoacetyl groups immo-

bilized on glass beads. Using this method, several phosphopep-tides were recovered from -casein and from a yeast cell ex-tract, although it was unclear whether all the proteins isolatedfrom the yeast extract were phosphoproteins (179).

Enrichment of the phosphoproteome can also be combinedwith protein profiling by 1- or 2-DE. In this way, changes inprotein amount observed on electrophoresis will reflect thelevel of protein phosphorylation (Fig. 13). Recently, the prin-ciple of protein quantitation by ICAT has been combined withphosphoprotein enrichment (60). This was accomplished bythe introduction of isotopic label into ethanedithiol, the re-agent used to convert the alkene created by -elimination ofphosphoserine into a free sulfhydryl group. In this way, the

differences in the amount of phosphoproteins in extracts canbe analyzed quantitatively in the mass spectrometer (60). Itshould be noted that because of the chemistry used in both ofthese methods, these techniques are relatively insensitive andrequire tens of picomoles of phosphoprotein. As a result, wehave found that these methods as currently designed are im-practical for the isolation and enrichment of low-abundancephosphoproteins.

Phosphorylation site determination by Edman degradation.

Edman sequencing is still a widely used method for determin-ing phosphorylation sites in proteins labeled with 32P, either in

vitro or in vivo (5, 22, 164). This is because sites can be deter-mined at the sub-femtomolar level if enough radioactivity can

be incorporated into the phosphoprotein of interest. In ourhands, this can be as little as 1,000 cpm (not ideal). Briefly, a32P-labeled protein is digested with a protease and the result-ing phosphopeptides are separated and purified by reverse-phase HPLC or thin-layer chromatography (TLC) (Fig. 14).The isolated peptides are then cross-linked via their C terminito an inert membrane (e.g. Immobilon P; PerSeptive Biosys-tems). The radioactive membrane is subjected to severalrounds of Edman cycles, and radioactivity is collected after thecleavage step. The released 32P is counted in a scintillationcounter. This method positionally places the phosphoaminoacid within the sequenced phosphopeptide. Of course, this ismeaningful only if the sequence of the phosphopeptide is al-ready known. In addition, the analysis ceases to become quan-



18/25

titative beyond 30 Edman cycles (even with efficient, modern

Edman machines) due to well-understood issues with repeti-tive yield associated with Edman chemistry.Recently, our laboratory has extended the usefulness of

phosphorylation site characterization by Edman chemistrythrough the development of the cleaved radioactive peptide(CRP) program (J. A. MacDonald, A. J. Mackay, W. R. Pear-son, and T. A. J. Haystead, submitted for publication). In CRPanalysis, one requires only that the sequence of the protein beknown. Purification and sequencing of individual peptides isnot required. Radiolabeled proteins (isolated following immu-noprecipitation from 32P-labeled cells, for example) arecleaved at predetermined residues by the action of a protease.The phosphopeptides are then separated by HPLC or TLC (ifonly one site is present, no peptide separation is required),

cross-linked to the inert membrane, and carried through 25 to

30 Edman cycles. The sequence of the target protein is enteredinto the CRP program. This program predicts how many Ed-man cycles are required to cover 100% of all the serines,threonines, and tyrosines from the site of cleavage. Generally,one round of CRP analysis narrows the number of possiblesites to 5 to 10 for most proteins. Phosphoamino acid analysiscan be used to reduce the number of possibilities still further.The CRP analysis is then repeated following cleavage with asecond protease (usually one cutting at R, but M and F arealternatives). The second round of CRP usually unambigu-ously localizes the phosphoamino acid to one possible site. Thetechnique does not work if sites are more than 30 amino acidsaway from all possible cleavage sites. The finding that CRPanalysis is not applicable may in itself confine a phosphoryla-

FIG. 13. Phosphopeptide and phosphoprotein enrichment. (A) Enrichment of phosphopeptides. Phosphoproteins are digested with a protease,and the phosphate groups are converted to

molecular biologist guide to proteo mics

Documents